The Effect of Reading Interventions among Poor Readers at a Forensic Psychiatric Clinic

Abstract

The aim of the present study was to investigate the effect of a short period, 15 sessions, of reading interventions in a sample of adult forensic psychiatric patients: 61 patients with decoding difficulties – 44 in the experimental group and 17 in the comparison group – with an average age of 31.6 participated. Of these, 36% were female, and 29% had an immigrant background. The participants carried out a battery of reading tests. The results in the experimental group showed a medium effect size (d = .36 to .76) on all reading tests between pre- and post-test. The comparison group, however, showed no gain at all between the test occasions. The results indicate that a proportionally low reading intervention effort produces improvement in reading. This study discusses the importance of including reading assessment and offering remediation in order to reach optimal future social adjustment for patients in forensic clinics.

Introduction

Adequate reading and writing ability is of major importance to manage daily life. Several studies have highlighted the importance of reading skills for successful schooling, whereas failure in school might enhance the risk of being marginalized (Archwamety & Katsiyannis, 2000; Grigorenko, 2006; Harris, Baltodano, Artiles, & Rutherford, 2006). Studies have shown connections between young offenders, reading disabilities, and future psychiatric illness (Carroll, Maughan, Goodman, & Meltzer, 2005; Eissa, 2010; Goldston et al., 2007; Mugnaini, Lassi, La Malfa, & Albertini, 2009; Teplin, Abram, McClelland, Dulcan, & Mericle, 2002).

A large number of investigations have shown that reading and writing disabilities and dyslexia are overrepresented (11–57%) among persons who have been committed to custodial care at juvenile institutions, prisons, and forensic psychiatric clinics (Svensson, Lundberg, & Jacobson, 2001; Moody et al., 2000; Selenius, Dåderman, Meurling, & Levander, 2006; Snowling, Adams, Bowyer-Crane, & Tobin, 2000; Svensson, Fälth, & Persson, 2015).

Researchers agree that learning disabilities (LDs) among inmates in juvenile institutions, prisoners, and patients at a forensic psychiatric clinic constitute an important factor that must be taken into consideration when planning and carrying out the treatment procedure (Blomberg, Bales, Mann, Piquero, & Berk, 2011; Dåderman, Lindgren, & Lidberg, 2004; Geib, Chapman, D'Amaddio, & Grigorenko, 2011; Selenius et al., 2006; Selenius, Hellström, & Belfrage, 2011; Svensson et al., 2015). Furthermore, the treatment procedure should include interventions based on the type of disabilities (Einat & Einat, 2008; Zhang, Barrett, Katsiyannis, & Yoon, 2011). Zhang et al. (2011) – for example, stress the importance of using an individual intervention strategy for offenders with different kinds of disabilities, ‘specialized interventions should also be provided to youth with disabilities while they are in juvenile facilities’ (p. 17).

Abundant studies have shown the importance of increasing reading ability among inmates in correctional settings (Svensson, 2011; Grigorenko, 2006; Selenius & Hellström, 2014; Selenius et al., 2006; Svensson et al., 2015). Rucklidge, McLean, and Bateup (2009) found that reading comprehension predicts, for example, the capturing rate, seriousness, and persistence of post-release offending – i.e., prison inmates with reading comprehension difficulties show more serious and more frequent offending. The authors also conclude that difficulties with reading and comprehension and LDs for these youngsters might jeopardize both educational interventions and other programmes and services at the institutions. Thus, quite a few studies have shown that educational programmes that include reading-related skills are associated with greater chances of managing different demands from society after a stay at correctional establishments such as prisons and juvenile institutions (for a review, see Vacca, 2008). Furthermore, studies by Vacca (2008) and Zhang et al. (2011) have established that educational programmes for juvenile offenders have a significant effect on the transition from custody to society and that the inclusion of reading skills might be one important factor in making this transition positive (Vacca, 2008).

The studies that have involved interventions concerning reading ability among incarcerated individuals have mostly been carried out among young offenders. Malmgren and Leone (2000) conducted a study where 45 incarcerated youngsters received a six-week intensive reading programme. The results show that their ability improved significantly in three out of four measured reading tests – namely, reading rate, reading accuracy and passage reading. However, the effect sizes between pre- and posttest among the tests were rather low (d = .12 to .17). The authors conclude that although a brief intervention can improve the reading performance of young people, their reading ability still remains low. For this reason, the authors state that longer-term literacy interventions are preferable in order to achieve more substantial differences among incarcerated youths. In a meta-analysis by Sander, Patall, Amoscato, Fisher, and Funk (2012) about the effect of interventions on academic outcomes among juvenile delinquents, the authors argued that the area lacks trustworthy information. Thus, there are few published studies, some of them with a weak design, such as lacking comparison groups and the use of best practice in interventions (Sander et al., 2012).

The aim of this study is to conduct reading interventions in the Swedish language for adult patients at a forensic clinic with a poor ability in reading Swedish. Few studies have conducted reading interventions with incarcerated individuals, and especially few, if any, have concerned psychiatric patients at forensic clinics. Even in the population at large the number of studies that have carried out interventions regarding literacy ability in adults is limited.

Reading Interventions among Adults

Torgerson, Porthouse, and Brooks (2005) did a systematic review regarding interventions in adult literacy. They concluded that since 1980 there had been only six evaluated interventions in literacy among adults where it was possible to calculate an effect size from the data presented. In addition, many of these interventions suffered from methodological problems, such as a lack of information about attrition, a lack of equivalence at baseline, and no reporting on all outcome measures as stated in the method part. The authors argued that it was difficult to draw any conclusions from these studies about the effectiveness of the interventions. Thus, the authors concluded that there was a need for more rigorously designed RCTs (Randomized Controlled Trials) and quasi-experiment studies of reading interventions among adults (Torgerson et al., 2005). A later RCT study by Greenberg et al. (2011) contained four reading-intervention groups: decoding and fluency; decoding, comprehension, and fluency; decoding, comprehension, fluency, and extensive reading; extensive reading; as well as a comparison group. Participants were 198 adults with reading equivalences between Grade 3.0 and Grade 5.9. The results showed that all participants gained significantly in their reading ability. However, the authors stressed that the effect size was low (d = .03 to .18), as had been shown in other, earlier studies (Greenberg et al., 2011). Furthermore, there were no differences between the four different intervention programmes. The authors advocated two main explanations of the non-significant results between different reading approaches. First, that the interventions are not suited for adult readers and, second, that the tests are not standardized for adult struggling readers. Consequently, we do not know whether these tests are sensitive to changes in different kinds of reading skills. In a study by Vanderberg, Pierce, and Disney (2011), the effect of reading interventions for adults with learning disabilities and low reading skills was examined. The results show that participants made moderate-to-large gains in reading accuracy (η² = .62) and comprehension (η² = .13), while there were no significant gains in reading rate. On the basis of their own study and a review of earlier publications, the authors raise the question whether the reading rate among adults with disability is at all possible to change. In a recent pilot study by Shore, Sabatini, Lentini, Holtzman, and McNeil (2015), 50 adults with a word recognition level below Grade 7 received 30 hours of reading fluency intervention. The results show significant gains in reading-related aspects after the intervention, a result interpreted by the authors as ‘highly encouraging’ (Shore et al., 2015, p 98). However, the effect size varied between low and moderate effects (d = .09 to .33), and this study lacked a control group. Thus, as Torgerson et al. (2005) stated in their review, there are few studies that give evidence of a specific intervention method that might be the most suitable for adult struggling readers. Furthermore, the effect sizes reported in the studies are often rather low with regard to gains in reading skills (Greenberg et al., 2011; Shore et al., 2015).

Patients at psychiatric forensic clinics

In Sweden there are researchers who have carried out reading and writing investigations among adult patients at psychiatric forensic clinics. All of them have drawn the conclusion that there is a huge over-representation of reading and writing disability in this group, and that it is essential to offer these patients reading and writing remediation at the clinic (Dåderman, Meurling, & Levander, 2012; Selenius et al., 2011; Svensson et al., 2015). However, it has not been possible to find any published studies, as far as we know, either in Sweden or in the rest of the world, that have carried out reading interventions among adult forensic psychiatric patients. Few of the institutions in Sweden have an established school activity, especially one with a focus on pedagogical efforts regarding written language, something that requires well-educated teachers with specific competence in the area. This is probably one reason why no investigation of reading intervention has been carried out (so far) among these patients in Sweden. Other reasons probably include methodological complexities. The existing conditions at these high-security institutions make it difficult to implement a RCT design, especially within a reasonable period of time and with reasonable economic resources. Ethical and security reasons often make it impossible to work with the patients in groups (Mulcahy, Krezmien, Leone, Houchins, & Baltodano, 2008). Furthermore, it is also quite common that patients do not want to work with other inmates because they do not want to reveal that they themselves have cognitive problems such as reading difficulties (information from the school staff at the institution participating in the present study). Therefore, a one-to-one design is often the only choice when carrying out intervention among this group of patients. However, this fact could actually be an advantage, since an individualized approach might be preferable. Furthermore, in a systematic review regarding harmful effects of crime intervention programmes, Brandon and Rocque (2014) found that, as harmful crime prevention programmes were more often conducted in groups, an individualized intervention programme was preferred.

Although an RCT design is preferable, the population of adult forensic psychiatric patients in Sweden is restricted because of the very few running systematic school activities, and even fewer experienced teachers who are able to carry out reading interventions at such institutions. Thus, it would take an unreasonable amount of time to collect a sufficient number of patients for an RCT design. Nevertheless, to our knowledge the reading level among these patients creates a need for qualified research about how to increase their reading skill. As these patients are a vulnerable group, increasing their reading ability will enhance their chances of rehabilitation into society, such as getting a job, restarting interrupted schooling, and decreasing recidivism. It might also increase the possibility of making their stay at the clinic more comprehensible and meaningful and to assimilate other programmes and activities at these institutions. (Grigorenko, 2006; Rucklidge et al., 2009; Svensson et al., 2015). A majority of this population has had bad schooling experiences involving failure in many subjects, not least as regards skills in reading and writing (Selenius & Hällström, 2014; Svensson et al., 2015). Some of the patients are almost afraid of school activities, since they remind them of previous emotional failures. It is also quite common that patients have difficulties with attention and concentration (Dåderman et al., 2004). These facts put high demands on finding suitable interventions for this group of patients. Thus, the interventions have to be meaningful for the patient and not too time-consuming. Furthermore, there are many reasons why the patients have difficulties in the Swedish written language, such as an immigrant background, bad school experiences, long periods of truancy, or dyslexic symptoms. These conditions point towards an individualized design, something that is also reinforced by earlier research (Greenberg et al., 2011; Torgerson et al., 2005) – i.e., there is weak evidence for using one single method to remediate reading among adults. Thus, a reading intervention programme among adults might benefit the most if it is individualized and includes issues such as decoding and reading comprehension. Furthermore, to increase the motivation to read, the text material should be based on their own interest (Greenberg et al., 2011). For this reason, Vanderberg et al. (2011) advocate a combined intervention approach comprising direct instruction in basic skills (bottom-up) and strategy instruction (top-down) with an emphasis on functional literacy for adult struggling readers. Thus, the present study involves a combined strategy of direct instruction of basic skills and strategy instruction, dependent on the participants' reading ability with an emphasis on functional reading ability – i.e., reading local newspapers, and having the confidence to write, for instance, postcards and letters to their children and parents.

The Present Study

In a recent study by Svensson et al. (2015), the reading ability of 185 patients from a forensic psychiatric clinic was investigated. The mean age was 30 years. Of the participants, 63% had a reading level, based on three reading-related tests (word-decoding, non-word reading and passage reading), that was below Grade 6. Especially severe difficulties appeared among male patients with an immigrant background and diagnoses of psychosis and anxiety. The conclusion was that the reading ability of one out of three patients was so low, especially with regard to reading comprehension, that it would probably jeopardize their everyday life, for instance when it comes to conversations with other patients and personnel, understanding text on television and in newspapers, or reading their records and institution rules (Svensson et al., 2015; Talbot & Riley, 2007). The low reading level found in that study is in line with several earlier (Alm & Andersson, 1997; Selenius & Hällström, 2014; Snowling et al., 2000). The present study takes the next step by conducting reading and writing interventions among 44 adult forensic psychiatric patients from the previous study (Svensson et al., 2015) who display reading problems in the Swedish language. The present study attempted to answer the following research questions: What are the general effects of an individualized intervention programme, with its aim to remediate reading skills, among adult forensic psychiatric patients with a low reading ability in the Swedish language? Are there any differences between gender, immigrants, and native-born Swedes or low and high performers regarding effects on the interventions?

Method

Participants

Sixty-one patients (37 men and 24 women) admitted to a forensic psychiatric centre in Sweden participated in the study. The clinic is fairly typical of this kind of institution in Sweden in terms of size and type of forensic patients, with the centre being classified as a high-security hospital with six different departments. The participants’ mean age was 31.6 years (SD = 9.2). Eighteen of the patients had an immigrant background, defined as having not being born in Sweden and both parents coming from a foreign country. Their residence in Sweden ranged from 6 to 49 years (M = 18.3, SD = 10.6), and all of them were able to speak Swedish. However, no assessment of their oral language ability in Swedish was carried out.

The target group consisted of forensic patients admitted to the clinic between January 2009 and May 2013. The patients in the study had different kinds of psychiatric disorders, divided into five categories: neurodevelopmental disorders (n = 17), mood disorders (n = 12), psychosis (n = 21), anxiety disorders (n = 6), and personality disorders (n = 3). Two of the patients were diagnosed with a mild intellectual disability. However, a non-verbal test shows that they achieved results in the normal range of the population, and they were therefore included in the study. The inclusion criterion for participation in the study was to have a result on a decoding test (Wordchains) and/or a non-word reading test (Non-word reading) that was below average for Grade 6 (age 12) in compulsory school. We regarded Grade 6 as a minimum reading level, as used in previous studies to manage ‘grown-up’ texts (Alm & Andersson, 1997; Sabatini, Shore, Holtzman, & Scarborough, 2011; Samuelsson, Gustavsson, Herkner, & Lundberg, 2000; Svensson et al., 2015). The reason for using a non-word reading test as well was that some adults might have to compensate for their decoding problems – i.e., they have a decoding ability almost in the normal distribution, which is not unusual, for example, among dyslexic adults. Thus, reading non-words will challenge the decoding ability more than reading ordinary words. Furthermore, before the first test session took place, a semi-structured interview was conducted with all participants, where they answered questions about their educational background and reading habits. All the participants in the experimental group and the comparison group belonged to the larger sample of 185 participants (Svensson et al., 2015). There were no significant differences regarding age or years in school between the full sample and the 61 adults who participated in the present study. Nor were there any differences regarding the distribution of psychiatric diagnoses between the samples. There were significantly more females (p < .05) in the present sample compared to the full sample. One reason for this difference was that females were more eager to participate in reading interventions than were males (for more details about the full sample, see Svensson et al., 2015). If the participants in the experimental group met the criterion, accepted participation, and were deemed by the doctor as sufficiently fit to participate in the investigation, they were asked to carry out reading interventions. The same procedure was used for the comparison group, except that those participants were not asked to carry out the reading interventions. Thus, 44 patients in the experimental group carried out the tests and interventions, while the comparison group of an additional 17 patients underwent the tests but not the intervention. Six patients declined participation, pleading that they did not want to go through the remediation procedure. Of the participants in the present study, 72% reported that they had some kind of reading and/or writing problems in the Swedish language as well as in their native language. The participants had completed approximately 10 years of formal schooling during childhood and adolescence. A majority of the participants (61%) reported troublesome schooling. The present investigation has received ethical approval (Reference number: 96/05) by the Ethic Review Board in Linköping, Sweden (EPN) (see Table 1).

• There were no significant differences regarding gender, age, years in school and language affiliation between the two groups.

Table 1.

Participants in the experimental and comparison groups presented by, gender, language affiliation, age, and years in school.

		Experimental group (n = 44) n (%) or M (SD)	Comparison group (n = 17) n (%) or M (SD)
Gender	Female	18 (40.9)	4 (23.5)
	Male	26 (59.1)	13 (76.5)
Language	Swedish	15 (34.1)	3 (17.6)
	Foreign	29 (65.9)	14 (82.4)
Age		30.5 (9.3)	34.5 (8.9)
Years in schoola		10.0 (3.1)	10.6 (1,5)

^aCompulsory schooling in Sweden comprises Grades 1–9.

Materials

The instruments used to measure the participants’ reading ability both for pre- and posttests cover decoding ability, phonological processing, orthographic ability (spelling), and reading comprehension. We also included RAN (Rapid Automatized Naming). The assessment battery included in this study contains aspects of reading that have been used by a great many studies when evaluating reading and writing disability and dyslexia (Lundberg, 2010; Rack, Snowling, & Olson, 1992; Samuelsson, Herkner, & Lundberg, 2003; Warmington, Stothard, & Snowling, 2013).

Wordchains

The test measures word recognition and involves discriminating three words in a chain without any space between them and then marking each word boundary with a pencil. The number of word chains marked correctly in 2 min was used as a measure of the participant's general word decoding skill. Standardized norms from Grade 1 to adult were available. The test-retest reliability is 0.92 (Jacobson, 1996).

Non-word Reading

A list of 84 pseudo-words consisting of 1–4 syllables each was presented to the participants. The task entails reading, in 1 min, as many non-words as possible from a list. Data for comparison were available (Grade 4 to Grade 9) and have been used in previous studies (Svensson, 2009; Svensson & Jacobson, 2006).

Phonological Choice

Pairs of pseudo-words were presented, with one word in the pair giving the impression of a real word whe34n sounded out, whereas the other word had no word likeness. The task was to decide which of the two words in a pair sounded like a real word. Pages of 20 word pairs were presented. The total working time was 2 min, and the performance was expressed as the number of correctly marked pseudo-words. Norms for the different age groups, Grade 3 to Grade 9, were available (Olofsson, 1994).

Orthographic Choice

The orthographic choice test consists of 130 real words paired with pseudo-homophones (e.g., cykel-sykkel, game-gaim), and the task was to decide which of the two was correctly spelled. Since both alternatives sound the same when pronounced, the decision has to be based on knowing the correct spelling. The number of correctly chosen words over a working period of 2 min was used as the performance score. Norms for the different age groups, Grade 3 to Grade 9, were available (Olofsson, 1994)

Non-word Repetition

The non-words were orally presented to the subject, and the task was to immediately repeat the word as correctly as possible. A non-word was considered as correct if it was repeated without any omission, substitution, or addition of sound. In order to include different degrees of difficulty, the test comprised some containing single consonants and others with consonant clusters. The test included 12 items. Comparison data from Grade 2 to Grade 9 were available (Fälth, Gustafson, Tjus, Heimann, & Svensson, 2013).

Passage Comprehension

A Swedish translation of a subtest from the Woodcock Reading Mastery Test–Revised (Woodcock, 1987) was used. After reading a text passage with a blank line, the students were instructed to fill in the blank orally with a single word that fit the passage. The first passages included a corresponding picture. Thus, the level of difficulty increased with the number of items, and the test ended when the participant had failed on six consecutive passages. The maximum score was 68. Data for comparison (Grades 2 to 6) were available from earlier publications (Svensson, 2009; Svensson, 2011).

RAN

Rapid automatized naming (RAN) of letters and numbers was assessed (36 items in each test). To test their reading fluency, the adults were to reproduce letters and digits rapidly and correctly (Norton & Wolf, 2012). The total time in seconds for naming both numbers and letters was used as a measure. Comparison data were available for Grade 6.

Interview

A semi-structured interview was used where the patients were asked questions regarding schooling, reading habits and their own perception of their reading and writing ability.

The answers were scored as ‘bad schooling or not’ and ‘reading and writing disabilities or not.’

Procedure

Test Procedure

All tests, in a one-to-one design, were administrated by two of the authors, who had a long experience of diagnosing and remediating persons with reading and writing disability and dyslexia. There were two pre-intervention tests (T1 and T2) intended to establish a baseline for each participant and one posttest (T3). The duration between T1 and T2 varied among the patients from 5 weeks up to 30 months (M = 7.6 months). The intervention took place between the second and the third test occasions, and T3 was carried out directly after the last intervention session. The duration between T2 and T3 varied from 5 to 6 weeks for both groups.

Intervention Procedure

There were three special educators (teachers with a university education in special education, hereafter named teachers), with several years’ experience of teaching in the reading area, who carried out all the intervention sessions. Although the teachers were familiar with the programme, they were instructed how to apply it with this specific group of participants. The participants received 15 one-to-one training sessions over a course of five weeks. It was not possible in this forensic psychiatric clinic to have any modalities other than a one-to-one design, for reasons of security and the doctor's decision about participation in the study. Furthermore, a majority of the patients did not want to work with other patients. There where no other school activities during the intervention period. The teachers were instructed to note in a diary the time they spent on the different aspects of the intervention as well as information about the patients’ mood and motivation when carrying out the session. A participant received, on average, 638 min (range = 455–712, SD = 51) training for the whole intervention period. On average they spent 310 min (SD = 83) with implicit (non-computerized) training and 337 min (SD = 72) with explicit (computerized) training. They carried out 13.6 sessions on average (SD = 0.9). There were no significant differences in the reading and spelling results after the intervention between the participants who received training above 637 min and patients who received training below 638 min.

The overall instruction for the interventions was as follows: in each intervention session participants chose and read, to themselves or via a text-to-speech program, an article from a daily web-based newspaper. It was important to choose texts to suit the participants’ own interests. After having ‘read’ the article, the participant had to write down, in a couple of sentences, the context of the chosen text. While writing the text, the participants had access to a spell-checker program (Stava Rex) especially adapted to persons with spelling difficulties. This part of the intervention was called ‘implicit training’. In addition to the instructions above, the teachers used two different computer programs aimed at giving individualized practice adapted to each participant's specific needs of having their reading and writing ability scaffold. This part of the intervention was called ‘explicit training’. The instruction to the teachers was to work with text-based training for half of the sessions and with the computer programs for the other half. The time spent was registered for the two parts of the interventions. Thus, these facts guarantee fairly good fidelity regarding the aims of the interventions. The teachers were told to individualize the sessions for each participant depending on their reading ability level – i.e., participants who had more fundamental reading difficulties used one program (Lexia) more than the others (Ordbygge). As the Ordbygge program requires a certain linguistic level in the user, the recommendation for the teacher was to use this program only if the tests had shown that this was the case.

Training Programmes

Lexia

Lexia (Gunnilstam & Mårtens, 2010) is a computer program for treating persons with dyslexia, aphasia, and other language-based learning disabilities. The program includes a large number of lists aimed at offering structural training for different language processes at all reading development stages. It is possible, for example, to practise linguistic and phonological awareness, vocabulary, reading comprehension, memory, sounds and letters on a phonetic and phonemic level, or language-based concepts. The program offers plenty of opportunities to practise each of the issues described above. One of the main ideas behind the program is to provide individual adjustment. The program is developed to adjust both younger and older pupils in the compulsory school.

‘Ordbygge 1, 2’ (creating words)

‘Ordbygge 1 and 2’ (Hadar, 1997) is an interactive and multimedial language-training program for practicing both reading and spelling. It is specially created for students in high school and for adults with reading and writing disabilities and dyslexia. The main purpose of the program is to stimulate linguistic and phonological awareness and give access to compensating strategies for reading and writing. In the program, the student can practise root phonemes, composite words, prefixes, suffixes and inflected morphemes. The program requires a certain linguistic level in the user, and is not recommended to use for younger pupils and for new immigrants who do not have adequate language skills.

Statistical Analyses

The first aim of the present study was to analyse the effect of an individualized reading programme and to compare the relation with a comparison group who received no interventions. Cronbach´s alpha was used to measure the agreement between scores at T1 and T2.

Longitudinal changes were analysed by paired sample t tests between the two pretests T1 and T2 as well as between the posttest T3 and the pretest T2. A change score (CS) between the posttest T3 and the pretest T2 was calculated for each reading test. An independent sample t test was used to reveal differences in pretest scores as well as posttest scores between the groups. When comparing the CS between the groups, a Welch t test was used. Furthermore, Cohen's d (Cohen, 1988) was used to estimate the effect of the reading intervention programme. Cohen's d was calculated as the ratio of the change score and the standard deviation of the mean score of the pretests in the pooled sample (all participants). Finally, a second aim was to examine possible differences in intervention gains among gender, immigrants, and native-born Swedes, and the impact of baseline performance (on pretests) as well as years in school in the experimental group. To this end, an independent samples t test or linear regression was used with a Bonferroni correction to account for the 8 reading tests.

Results

There was a strong agreement between the test scores at T1 and T2 (Cronbach´s alpha of .95 or above for all variables). Pre-reading values and change scores between T1 and T2 are presented in Table 2. At the pretest, the comparison group had slightly better results than did the experimental group on all the tests (Wordchains, Pseudo-word-reading, Phonological choice, Non-word repetition, Orthographic choice, Passage comprehension, RAN letter, and RAN digits), though the differences were not significant, except for pseudo word reading (p < .05). When a non-parametric test, Wilcoxon Signed Ranks test, was used, none of the tests was significant between the groups. Furthermore, the paired t tests showed no significant differences between the two pre-tests.

Table 2.

Change scores between T2 and T1 for the whole sample.

	Both groups (n = 61)		Experimental group (n = 44)		Comparison group (n = 17)			Norms Grade 6
	Tests	CS	pa	M	SD	M	SD	pb	M	SD
Wordchains	–.28	.28	27.5	12.8	28.1	8.3	.86	36.9	7.8
Pseudo word reading	–.52	.14	21.8	10.0	28.9	12.4	.02*	37.5	10.7
Phonological choice	–.16	.54	12.1	7.3	13.9	9.9	.43	24.7	7.9
Non-word repetition	.00	1.0	8.0	2.3	8.1	1.9	.91	9	1.6
Orthographic choice	–.20	.69	47.0	25.3	54.6	22.2	.29	59.5	20.5
Passage comprehension	–.15	.67	25.5	12.7	29.6	12.3	.26	39.1	7.6
RAN Letter	.69	.06	19.4	6.1	19.9	7.1	.81	14.1	2.5
RAN Digit	.34	.21	17.8	4.1	17.4	7.6	.76	15.6	1.8

Notes: Whole sample: n = 61. Mean raw scores for the reading, spelling and the RAN tests for the pre-test occasion T2. Comparison data for Grade 6. CS = Change score (T2-T1).

^aPaired t test between T2 and T1, ^bIndependent t test between experimental and comparison group.

The posttest gave the opposite result (see Table 3) – i.e., the experimental group had a higher score on all reading and spelling tests, still without significant differences. However, the experimental group had significantly higher change scores on all reading and spelling tests compared to the comparison group (p < .001).

Table 3.

Mean raw scores at T3, change scores between T3 and T2, and effect sizes for the reading, spelling, and RAN tests.

	Experimental group (n = 44)			Comparison group (n = 17)			Effect size
	Tests	M	SD	CS	M	SD	CS	d	CI	p
Wordchains	33.3	13.6	5.8	28.6	8.2	0.3	.43	[.33, .52]	<.001
Pseudo word reading	30.0	13.8	8.2	28.5	11.3	−0.4	.66	[.47, .84]	<.001
Phonological choice	20.0	11.5	7.8	14.9	11.2	0.9	.60	[.40, .81]	<.001
Non-word repetition	9.5	2.1	1.4	8.2	1.9	0.1	.67	[.47, .87]	<.001
Orthographic choice	57.7	28.7	10.7	55.5	22.0	1.0	.36	[.22, .50]	<.001
Passage comprehension	38.2	16.4	12.7	30.5	13.0	0.9	0.76	[.62, .90]	<.001
RAN Letter	17.9a	5.4	−1.8	19.5	6.7	−0.4	−.24	[−.45, .02]	0.03
RAN Digit	16.7a	3.9	−1.3	17.5	7.5	0.3	−.32	[−.48, .16]	<.001

Notes: CS = Change score; d = Cohen´s d between posttests and pretests; p = Welch t test of the CS between the experimental and the comparison group.

^aLower scores indicates better results.

The participants’ pretest score on the reading tests was approximately equivalent to Grade 4 for the comparison group and below Grade 4 for the experimental group on the basis of comparison data and standardized tests. At the posttest the comparison group had the same level on the reading tests as at the pretest, while the participants in the experimental group had raised their scores on the reading and spelling tests to the average for, or above, Grade 4. On the RAN tests, the experimental group surpassed the comparison group.

The pretest–posttest effect size (Cohen's d) in the experimental group varied from d = .36 to d = .76 on the different reading and spelling tests – i.e., a medium effect. In the comparison group the effect was very small or there was none at all. Reading comprehension (passage comprehension) showed the highest gain (d = .76) between the two test occasions. Figure 1 contains data from the reading comprehension test at pretest and posttest. The scatter plot shows that all of the participants in the experimental group increased their ability on this test, whereas most of the participants in the comparison group showed practically no difference at all.

Figure 1.

A scatter-plot for the results on Passage comprehension (reading comprehension) between pre- and posttest in the two groups.

Differences in Gains between Pretests and Posttest for Women and Men, for Swedes, and for those with a Foreign First Language and for Low and High Performers

A second aim in the present study was to reveal possible differences in gains after the interventions between Swedes and those with a foreign first language, between women and men, between high and low performers (on pretests) and years in school. After correcting for multiple testing, the change scores differed only between genders. Females had significantly higher gains on passage comprehension (17.1 vs 9.8, p_c = 0.001), phonological choice (11.5 vs 5.3, p_c = .02), and orthographic choice (16.5 vs 6.7, p_c = 0.03). However, for the males in the intervention group these gains were all significantly better (p < .005) than in the comparison group.

Discussion

The aim of this study was to investigate the effect of a short period of reading and writing interventions among forensic patients who were perceived to be below Grade 6 on reading tests. All patients who carried out the interventions increased their ability on word decoding, phonological decoding, spelling, and reading comprehension. This was most obvious among women.

In the present study the patients’ average level on the reading tests was equal to Grade 4 or below. This is alarmingly low and will most certainly affect their everyday life at the institution, including reading and understanding texts in newspapers, their own records, and the institution rules. After the intervention, the experimental group in the present study increased their ability on all reading outcomes by an average effect size of d = .58 (range = .36 to .76). In a clinical perspective this means an increased ability of almost two grades on a group level. Still, the patients are only at approximately Grade 5 level. Nevertheless, even a limited improvement has a potential impact – for example, when applying for a job – and could have a functional effect (Vanderberg et al., 2011). For some of the poorest readers in the present study it was obvious that the reading intervention had a functional effect – i.e., they started to read and write postcards to their family and to read their local newspaper, which extended the conversation with family members to include topics about their neighbourhoods.

In the study by Vanderberg et al. (2011), a moderate- to large effect was found on reading accuracy and on reading comprehension, but no gains in reading rate. As the authors did not find any gains in decoding speed, they raised the question whether it is possible to increase decoding ability among adults. In the present study we found gains with medium effect size (see Table 3) in decoding ability, including speed, both in word recognition (Wordchains) and phonological decoding (non-word reading and phonological choice). The differences between the present study and earlier studies regarding decoding speed might be due to the severity of the participants’ reading problems and their frequency of learning disabilities. Since the present study did not have the aim to select exclusively those with severe reading difficulties, dyslexia, or other learning disabilities, it probably did not include a large number of participants who displayed these specific markers. Thus, adult individuals with markers of severe reading difficulties and learning disabilities might be more resistant to interventions. Another explanation of discrepancies between the present study and earlier studies might be differences in orthographies between languages. In Swedish accuracy is not as important as it is in English, as it is a more transparent language. It may therefore be easier to enhance the reading rate in Swedish even with a rather limited effort.

The participants in the present study made the greatest gains in a reading comprehension test (passage comprehension, d = .76). The ultimate goal of reading is, of course, to understand the message of a text. Still, even if you are an accurate and fast decoder, this is not enough to understand the text. In a study by Rucklidge et al. (2009) the authors state that poor reading comprehension is a factor predicting recidivism. They argued that poor comprehension was predictive of future post-release offending with regard to its rate, seriousness, and persistence. It is understandable that poor reading comprehension is an important factor to make sense of everyday life – not only of written text, but also of conversational talk. This may even be the case for adults with a fairly good vocabulary in an oral context the first time they hear a word. ‘Their vocabulary may lack the semantic depth typically acquired through multiple, varied exposures in text’ (Vanderberg et al., 2011, p. 212). For participants in the present study who improved their reading comprehension this may also increase their ability to better understand conversations of different kinds that go on between patients or between patients and psychiatric nurses, as well as in different kinds of therapeutic treatments (Svensson et al., 2015).

Literacy difficulties among the patients in the present study are most likely due to several reasons such as limitations in the Swedish language for those with an immigrant background, deficiencies that might have a dyslexic origin, poor schooling, and other literacy difficulties due to the participants' psychiatric diagnosis. Reading problems among adults often have multifactorial causes that indicate a need for a wider range of interventions (Mellard & Fall, 2012; Sabatini et al., 2011). The interventions in the present study used programmes with a quite broad area of reading exercises in order to adjust the achievement as much as possible for each participant.

The improvements in the present study were visible even among subgroups, such as difference in gender, for Swedes and for those with a foreign first language, and for high and low performers (everyone in the experimental group had a low reading ability compared to typical readers). It was most obvious among women who, on a group level, actually did not present any difficulties in reading after the intervention period. Their scores approximated the grade equivalents of Grades 8 and 9. The fact that the women made the most obvious improvements might be due to their self-efficacy with regard to literacy issues. They reported bad schooling and looked upon themselves as low performers in almost every school subject. This short time of intervention was, however, enough to ‘wake them up from a slumbering ability’. It was also obvious, when recruiting the participants and during the intervention sessions, that the females were more motivated to carry out a study with the aim to increase their reading and writing ability. Even if females had significantly higher gains than males, the males increased their ability on all the reading tests compared to the comparison group (p < .005).

Furthermore, the results showed no differences in gains between low and high performers. Neither was there any difference between participants with an immigrant background and Swedes. Thus, an individualized intervention approach seems to have the same effect despite various reasons for difficulties in reading in the Swedish language.

Most of the participants in the present study probably do not have dyslexia. Dyslexics often gain slowly from reading remediation and thus require a remediation programme over a much longer time than the patient in the present study obtained to receive any substantial effect. The same is also true for participants with a too limited knowledge in the Swedish language, i.e., the intervention period has to last much longer and include more training sessions to reach at least Grade 6 levels in reading and spelling. Even if the immigrants gained as much as the Swedes, the immigrant patients’ reading ability in Swedish was as low as Grade 4 in the present study. It is not hard to imagine the frustration these patients must feel in their everyday life at the institution, not least when trying to express their thoughts and feelings. Nevertheless, even though males’ and immigrants’ pretest ability was very low and remained so after the posttest, their gain in reading ability was promising, which underlines the possibility as well as the necessity of remediating the reading ability for these groups. In fact, the most promising finding in this study may be that this adult sample, including different subgroups with varied reading difficulties, did show a substantial response to intervention.

As the time for the intervention was on average only about eleven hours (13.6 sessions), the improvement in the present study could be regarded as promising, at least when comparing to other reading intervention studies among adults, whose results show lower gains and/or substantially more time spent on interventions (Sabatini et al., 2011; Vanderberg et al., 2011). For example, in Vanderbergs et al.'s study (2011) the participants received, on average, 85 hours of interventions. Possible reasons why the outcomes showed rather high gains in this study despite shorter periods of interventions, compared to some other studies, might be due to the intervention approach, i.e., the fact that they were carried out in one-to-one tutoring sessions. An additional possible reason was that the teachers were successful, in most cases, in creating a good secure pedagogical situation. Several of the patients said that this was the first time or one of the first times they had perceived a school situation as positive. Many of the participants, especially the females, perceive participation in this study as motivating and meaningful. The impact of the close relation between the patient and the intervener during the sessions is probably not negligible in the light of an increasing motivation. The teachers reported that several of the patients who carried out the interventions made the comment that they never had worked so intensively with a school issue before.

Within the duration of the present study, none of the participants were engaged in any other school-related activities than this reading intervention for the experimental group. In fact, for most of the participants the intervention sessions were the only activity they were offered during this period. Thus, we may argue that the gains observed in this investigation are most likely due to the reading interventions and not to other school activities or to maturity. In the comparison group there were no significant gains on the test results between pretest and posttest, which strengthens the power of the intervention effect in the experimental group. However, as in many one-to-one interventions it is tricky to sort out what is the main reason for the gain, i.e., is it the way the intervener acts in the one-to-one-teaching situation that might increase the motivation for the participants, is it the specific training programme that has been used or is it a combination? Nevertheless, the short time of reading interventions that was offered in the present study has obviously increased the ability in reading that may enhance the feeling of meaningfulness during the time the participants are staying at institutions of this kind.

In the present study the aim has not been to compare intervention methods between groups with a low decoding ability. Instead, the teachers had individualized the interventions depending on the participants’ main problem with reading or writing. We found this approach more relevant in this setting since there are many reasons behind their low decoding ability. Earlier studies have shown that there are no significant differences between varieties of reading remediation methods among adults with reading difficulties (Greenberg et al., 2011; Torgerson et al., 2005). Furthermore, in a recently published review, Elliott and Grigorenko (2014) argued that there is no obvious evidence that one reading method is superior to another, at least not when comparing interventions among those with a dyslexic diagnosis and those with general reading disabilities. The result in the present study confirmed that assumption, since the participants received different kinds of interventions depending on what kind and severity of problems they presented, and there were still no differences in gains among different subgroups. As mentioned before, there are few, if any, studies that have carried out reading interventions among forensic patients. It has therefore not been possible to compare the results in the present study with earlier investigations that have used the same population. However, the participants in the present study made obvious gains on the reading outcomes, which in a clinical perspective is important in order to be prepared for the literacy demands that are required in society after their stay at the forensic psychiatric care. Even during their stay at the clinic it is important to improve language skills in general, and reading skills in particular, to make the treatment procedure more meaningful and intelligible. Furthermore, there is consensus regarding correlations between inability to read, delinquency and recidivism among researchers (Katsiyannis, Ryan, Zhang, & Spann, 2008). Studies that have carried out reading interventions among incarcerated youths have shown several benefits with an increasing literacy ability, such as a reduced risk for recurrence in criminality and an increasing possibility to find a job and to continue interrupted schooling (Grigorenko, 2006; Mulcahy et al., 2008). Even if this has not yet been confirmed among patients at a forensic clinic by earlier studies it may be possible that this assumption is eligible too, among the participants who carried out reading interventions in the present study, after their stay at the clinic. However, more comprehensive and longitudinal research has to be carried out which includes, for example, longer periods of reading interventions, a follow-up regarding the long time effect of the interventions, the effect on every day life, schooling, getting jobs and recidivism.

Limitations

There are limitations in the present study. First, the participants were not randomized into experimental and comparison groups. This was due to the specific sample where we had to prioritize those that wanted to take part in the intervention period. Consequently, there were not enough patients for carrying out the study within reasonable time, especially since some of them refused to undergo the interventions. A second limitation is that the comparison group included quite a few participants with different initial results on the reading tests compared to the experimental group. A procedure with matching the initial reading ability would have been preferable but, due to the low number of participants in the comparison group, that was not possible. However, there was no gain whatsoever in the experimental group or the comparison group between the two pretest occasions, and there was no gain at all between pre- and posttests in the comparison group. These facts are evidence that the reading interventions for the experimental group have affected the participants' reading level in a positive way. Thus, there is a gain on the reading measures after the intervention for the experimental group. However, a majority of the participants in this group have still not reached the Grade 6 level on several reading tests. For the patients in the present study, with severe difficulties or a low knowledge in the Swedish language, it most likely demands a considerably larger effort to reach an acceptable reading level. A third limitation is that several of the tests have no standardized norms for adults (Greenberg et al., 2011). This is an overall problem regarding Swedish reading tests for adults and makes the interpretation of the test results somewhat unsteady (Swedish agency for health technology and assessment of social service [SBU], 2014).

Conclusion

A large number of investigations show an obvious overrepresentation of reading disabilities (50-70%) among inmates and patients in juvenile institutions, prisons and forensic psychiatric clinics. These studies often argue for the importance of including assessments of reading ability in the overall investigation, as well as remediating reading problems among those with reading difficulties. The remediation of reading problems is a crucial factor for decreasing recidivism into criminality as well as increasing the ability to continue previously interrupted schooling and receive a job (Archwamety & Katsiyannis, 2000; Grigorenko, 2006; Harris et al., 2006). However, among adults, reading intervention studies are fairly scarce, and among patients in a forensic clinic they are almost non-existing, as far as we know. The present study has shown that a short reading intervention period brings improvement in a group of forensic patients with a low reading ability. Since reading proficiency seems to have a strong predictive factor for, e.g., recidivism or obtaining a job, it will be important for institutions like forensic psychiatric clinics to include remediation that focuses on improving the reading ability and particularly reading comprehension in patients’ overall treatment plan to make the transition from custody to society a positive one. Actually, there is a unique opportunity to carry out interventions of different kinds, such as reading programmes, because of the lack of daily ongoing activities at this type of clinics.

Acknowledgements

This research was supported by a grant from Kronoberg County Council, Sweden.

Disclosure statement

No potential conflict of interest was reported by the authors.