The Association Between Fear of Crime, Life Satisfaction, and Health-Related Quality of Life in Non-Victimized Older Adults Aged 60–93 Years—Findings from the Swedish Good Aging in Skåne (GÅS) Population Based Study
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
This paper could benefit from the comments listed in the attached file.
Comments for author File: 
Comments.pdf
Author Response
International Journal of Environmental Research and Public Health April 24, 2026
Cover Letter
Response to Reviewers
Manuscript ID: ijerph-4195905
Type of manuscript: Original Article
Title: The Association Between Fear of Crime, Life Satisfaction, and Health Related Quality of Life in Non-Victimized Older Adults Aged 60-93 Years. Findings From the Swedish Good Aging in Skåne (GÅS) Population Based Study.
E-mail: henrik.ekstrom@med.lu.se
Dear Editor,
Please find enclosed the revised version of our manuscript with the title “The Association Between Fear of Crime, Life Satisfaction, and Health Related Quality of Life in Non-Victimized Older Adults Aged 60-93 Years. Findings From the Swedish Good Aging in Skåne (GÅS) Population Based Study.”
Manuscript ID: ijerph-4195905
We are grateful for the opportunity to improve the manuscript and we want to thank the reviewers for their fruitful comments and suggestions, which will be responded to by the following point-to-point replies. We hope that the revisions and responses will be sufficient for our manuscript to be accepted for publication in the International Journal of Environmental Research and Public Health.
All authors have approved the revised and re-submitted manuscript.
Henrik Ekström
Reviewer 1
A review of “The association between fear of crime, life satisfaction, and health related quality of
life in non-victimized older adults aged 60-93 years. Findings from the Swedish Good
Aging in Skåne (GÅS) population based study.”
This study investigates the association between behavioral fear of crime and the quality of
life among older adults aged 60-93 by examining various aspects of the quality of life,
physical and mental characteristics, and life satisfaction adjusted for demographic,
socioeconomic, and health-related covariates. The authors are using a well-known
population-based aging platform in Sweden, drawing a cross-sectional sample from waves
of a longitudinal study.
The use of multiple linear regression appears appropriate given the continuous nature of the outcome variables (LSI-A, SF-12 PCS and MCS). However, the manuscript would benefit from clarification regarding assessment of regression assumptions and whether robust standard errors were considered. Additionally, several covariates (e.g., depression, ADL limitations, physical activity) may cause fear of crime and quality of life. The authors are encouraged to clarify their conceptual framework and discuss whether these variables are treated as potential mediators. Finally, given the large sample size, greater attention to the substantive (not only statistical) magnitude of effects would strengthen interpretation. More detailed comments are listed below.
General Comments
Comment 1.
The manuscript would benefit from thorough language editing. There are numerous grammatical and lexical issues (see specific comments below). I strongly
recommend professional language editing or careful proofreading by a native
English speaker before resubmission.
Authors’ reply
The manuscript has undergone a linguistic review by English speakers and Swedish speakers, where the text has been translated between the languages ​​a couple of times.
Comment 2.
There are certain inconsistencies. For example, the authors say: 11,652 invited
(earlier), then 11,562 invited (later). That discrepancy (90 people) needs clarification. Also, “remaining 10,474 eligible informants”, but later: “Finally, out of 10,747 eligible individuals…” This is not fatal, but numbers should be adjusted for consistency.
Authors’ reply
The number and proportions of participants, both in total and in different subgroups, are now corrected both in text and figure.
Comment 3.
In section 2.2 the authors report collapsing five categories into four by merging
“quite often” with “often” However, it is not clear whether this decision was prespecified or data-driven. If the authors at the distribution and then merged based on frequencies, that’s fine — but it must be transparently justified.
Authors’ reply
We want to try to clarify this. The reason for merging the answer options "quite often" and "often" has now been clarified by the following sentence, line 141-144.
“The responses Quite often and Often represented only 5.9% and 4.1% respectively and were therefore combined into a single category labeled Often, resulting in a variable with four response options.”
Comment 4.
Lines 194-195: “All the presented independent variables were simultaneously
entered into the regression models and all variables were used as dummies.” This sentence is imprecise because questions immediately arise: was FOC entered as a categorical variable with 3 dummy variables? Or was it treated as ordinal or continuous? Did the authors dummy-code all predictors, including continuous ones? The earlier list of variables includes age, financial status, physical activity, ADL, and cognitive status. Some of these variables are likely continuous, so “dummifying” them seems odd. Please make this part of the discussion more transparent.
Authors’ reply
We agree that this approach may seem a bit strange, but we have tried to clarify this. Line 215-217.
The sentence now reads: " All the presented independent variables and the dependent variable FOC were simultaneously entered into the regression models and both the independent variables and the dependent variable FOC were used as dummies.”.
In the strengths and weaknesses section, we have also explained the choice of dummies. Line 414-418.
“Another possible limitation is that we categorized continuous variables in our regression models, which often results in a loss of information. The reason was that we wanted to make it easier to interpret the results. It is often easier to discuss, for example, what a dependency in ADL or a depressive state means in relation to FOC, than to interpret what a change in a continuous scale means.”
Comment 5 a.
Regarding the multivariate regression analysis in Section 3:a. The authors should clarify whether regression assumptions in Section 3.2 (e.g., normality of residuals, homoscedasticity) were assessed and whether robust standard errors were considered.
Authors’ reply
In the statistics section we have clarified this, it now says. “In each multiple regression model normality was controlled for by inspecting histograms of the residuals, and linearity by checking scatter plots (standardized predicted values vs. standardized residuals) but no unacceptable deviations were noted [27]. To address heteroscedasticity, the results of the regression models are presented with calculated robust standard errors (HC3). Sensitivity analyses with multiple imputations of missing data were conducted. Five imputed datasets were generated using the fully conditional specification method (FCS). The imputed model included all variables used in the regression analyzes. Parameter estimates were pooled according to Rubin's rules (Supplementary tables 1-3).” Line 220-229.
Comment 5 b.
Several covariates (e.g., depression, ADL, physical activity) may lie on the causal pathway between fear of crime and quality of life. The authors may wish to clarify their conceptual model and whether these variables are treated as confounders or potential mediators.
Authors’ reply
We have tried to clarify this by adding the following lines to the introduction:
“The theoretical framework we have used in this study is how a society's perceived crime subjectively and emotionally (affectively and cognitively) can lead to FOC shaped by an individual's previous experiences of victimization, environmental disorder, physical and psychological vulnerability or poorer social integration [8] and whether FOC in turn can affect QoL, either directly or via changed behavior. The relationship we have analyzed is whether FOC expressed as a changed behavior, not daring to go out in the evening for fear of being exposed to violence or threats of violence, and whether this fear is associated with QoL adjusted for lifestyle, economic, medical and social covariates.” Line 78-85.
Comment 5 c. Given the large sample size, the authors may consider discussing the clinical
or substantive significance of the observed differences in addition to statistical significance as, given n ≈ 5,500, everything will be statistically significant.
Authors’ reply
The magnitude and practical meaning of the associations between FOC and QoL has been shortly commented on in the second paragraph of the discussion with the following text:
“Although we have shown statistical relationships between FOC and QoL, the question is whether these are relevant and whether the results mean anything in practice. In general, it can be said that it depends on what the study population looks like and in what context the survey is conducted. However, if one looks at life satisfaction as a global measure of well-being, the norm values for a general Swedish population 60 years and older have been calculated for the LSA-A to be 28 points with a standard deviation of 7 points. The adjusted difference of 2 points, between the groups of never being influenced by FOC in their behavior to very often doing so, is not insignificant [18]. In addition to a worse mood and deteriorating mental and physical health and a reduced social engagement of the individual [28], lower life satisfaction can indirectly be a cause for greater FOC. That is, the inverse relationship we have tried to highlight, and a vicious circle between FOC and QoL can become the case, possibly even more contribute to negative societal consequences, for example an increased burden on healthcare and social care, both in terms of work effort and costs [29].”, Line 309-322
Comment 5 d.
The cross-sectional design limits causal inference. Wording should reflect associations rather than directional effects.
Authors’ reply
We have corrected the manuscript regarding inappropriate wording that expresses any form of causality and instead write about associations/relationships.
Comment 6.
In the Discussion section the authors talk about the FOC paradox, which is off-topic given the fact that in their multivariate regression analysis FOC is always an independent, rather than the dependent, variable. The authors may wish to clarify how this discussion contributes to interpreting the main findings. The authors say that “FOC paradox was confirmed in this study…”, however, they don’t model FOC, which makes their claim descriptively consistent at best.
Authors’ reply
Thank you for pointing this out, we agree that we expressed ourselves inappropriately. At the same time, we think there may be some interest in keeping a few lines about the FOC paradox. We have tried to work around the problem by rewording the text slightly with the following sentence.
".. As mentioned above, the FOC paradox [6], i.e., those with the strongest fear of becoming a victim of crime are those least affected by violence or threats of violence, was in line with the descriptive findings from this study. Women and participants aged 70 and older reported higher levels of FOC than men and 60-year-olds [30, 31]..." Line 323-327.
Comment 7.
The discussion (lines 298–317) acknowledges potential bidirectional and reinforcing relationships between fear of crime, mental health, and quality of life. However, these dynamics imply possible reverse causation and residual confounding in the estimated associations. Given the cross-sectional design, the direction of effects cannot be established. The manuscript would benefit from more explicit acknowledgment of this limitation and from consistently framing the findings in terms of associations rather than causal effects.
Authors’ reply
Yes, we of course agree with this, we have reviewed the manuscript and changed the sentences that suggest causality to "associations". To emphasize that we cannot speak of causality, we have added the following sentence to the strengths and weaknesses section:
“However, it should be said that the cross-sectional design of this study precludes the evaluation of the direction of the associations of FOC and LS and HRQoL.” Line 425-427
Specific Comments
Comment 1.
Capitalize words in the title
Authors’ reply
The title is corrected regarding capitalization
Comment 2.
Line 49: The conjunction “but” suggests a logical contrast between “unpleasant”
and “positive function,” which are not conceptually opposed. Consider replacing it with “and” or restructuring the sentence. Similarly, the conjunction “but” in line 55 suggests a contrast between being less commonly described and having serious consequences; however, these properties are not conceptually opposed. Consider restructuring the sentence to improve parallelism and clarity.
Authors’ reply
These linguistic (logical) errors have been corrected; "but" has been replaced with "and", line 50 and 57.
Comment 3.
Line 73: “and increasingly heterogenous”→ and becomes increasingly Heterogeneous
Authors’ reply
The sentence has been corrected to “However, the older population is growing and becomes increasingly…”, line 75.
Comment 4.
Line 83: get rid of the semicolon, e.g. make two sentences
Authors’ reply
The semicolon has been replaced with a comma. Line 87.
Comment 5.
Line 91: “has been described previously”→is described in
Authors’ reply
Corrected, it now says: “The design of the GÅS study is described by Ekström et al. [14].” Line 109.
Comment 6.
Lines 91-93: The sentence should read like “Individuals randomly selected from the national population register were invited by letter or phone. Informed consent was
obtained from those who agreed to participate.” I don’t insist the authors copy my wording, just saying the original sentence is awkward and should be improved.
Authors’ reply
We copied your suggestion, line 110-112. We really appreciate your corrections, the poor English is a problem for us, yet English speaking employees have read the script and approved it. Maybe they have lived in Sweden too long?
Comment 7.
Line 132: ranges from→ranges between
Authors’ reply
Corrected, line 152.
Comment 8. Line 145: has been→have been
Authors’ reply
Corrected, line 165
Comment 9.
Line 206: Figure 1→Table 1?
Authors’ reply
“Fig. 1.” Has been replaced by “Table 1.”, line 240.
Comment 10.
Tables 2a and 2b should be better referred to as Tables 2 and 3
Authors’ reply
Tables renamed, table 2a is now table 2 and table 2b is table 3.
Comment 11.
Line 239: what does Table 2 mean? Is it 2a or 2b? The discussion in lines 236-239 should be moved to section 3.1
Authors’ reply
This has now been corrected. The text lines previous 236-239 have been moved to section 3.1., line 251-254. “Table 2” replaced by “Tables 2 and 3”, line 254.
Comment 12.
In reporting their regression results in Section 3 the authors are advised to report confidence intervals below the coefficient values, which improves the paper’s
Readability.
Authors’ reply
With all due respect, we are grateful for this recommendation but would like to keep the tables as they are now. The reason is that confidence intervals under the odds ratios create a certain mess in the tables. Graphically, it does not look good.
Comment 13.
Lines 274-275: “greater proportion of the oldest old and women reported FOC compared to men and younger people” is incorrect
Authors’ reply
We have reformulated the sentence, it now reads: "Women and participants aged 70 and older reported significantly higher levels of FOC compared to men and participants aged 60." Line 325-327.
Reviewer 2 Report
Comments and Suggestions for Authors
This study investigates the association between behavioral fear of crime (FOC) and health-related quality of life (HRQoL) as well as life satisfaction (LS) in a large sample of non-victimized older adults drawn from the Swedish GÅS population study. The topic is timely and clinically relevant, and the exclusion of previously victimized individuals is a meaningful design choice that strengthens the focus on the FOC paradox. The large sample size, population-based design, and the inclusion of chronic disease categories as covariates are notable strengths. However, several methodological and reporting issues require attention before the manuscript is suitable for publication.
1. there are multiple numerical inconsistencies in the participant flow description that must be resolved. Line 93 reports that 11,652 individuals were invited across four waves, whereas line 100 states that 11,562 GÅS informants were invited — a discrepancy of 90 individuals that is not explained. Similarly, the wave 3 sample size is reported as n = 2,012 in the text (line 97) but as n = 2,018 in the flow diagram (Figure 1). Most importantly, line 106 states that the study sample comprised 5,832 participants "out of 10,747 eligible individuals," but the preceding arithmetic yields 10,474 eligible individuals after exclusions (11,562 minus the 1,088 who were not contactable, had moved, had language difficulties, or had died). The reported participation percentage of 55.7% is consistent with a denominator of 10,474, not 10,747. These discrepancies undermine confidence in the data pipeline and must be carefully verified and corrected throughout the text and figure.
2. The pooling of data from four assessment waves spanning over two decades (2001–2022) without adjusting for wave or calendar period is a substantial methodological concern. Societal factors that plausibly influence both FOC and quality of life — including changes in crime rates, media coverage of violence, urban development, and the COVID-19 pandemic — changed substantially during this period. The authors briefly mention cohort effects in the limitations (lines 370–374), but this acknowledgment does not address the analytical gap. At a minimum, the regression models should include wave or assessment period as a covariate, or a sensitivity analysis should demonstrate that the associations are consistent across waves. Without such adjustment, it is difficult to disentangle the FOC–QoL association from secular trends.
3. The handling of missing data is not described. The regression model sample sizes differ across outcomes (n = 5,373 for LSI-A, n = 5,367 for PCS, and n = 5,519 for MCS), suggesting that listwise deletion was applied. Given that MADRS has 6.7% missing values and MMSE has 3.8%, the cumulative effect of missing covariates may be non-trivial. The authors should state explicitly how missing data were handled, report the percentage of complete cases used in each model, and discuss whether the missingness pattern could introduce bias. Sensitivity analyses using multiple imputation would strengthen the findings.
4. The dose-response pattern for FOC and mental HRQoL (MCS) is not monotonic: the "Very often" category (B = −3.17) shows a smaller coefficient than the "Often" category (B = −3.75) in Table 5, which breaks the graded pattern observed in the LSI-A and PCS models. This irregularity deserves explicit discussion. Possible explanations include the small size of the "Very often" subgroup (n = 258, 4.4%), ceiling or floor effects in MCS scoring, or unmeasured characteristics of those with the most extreme avoidance behavior. The authors should not present a uniform dose-response narrative without addressing this deviation.
5. The Highlights section and the Conclusion contain causal language that is inconsistent with a cross-sectional design. For example, line 17 states that "Fear of crime is a stress factor leading to physical and mental illness," and lines 26–27 and 386 state that "such fear can lead to a reduced quality of life." Although the authors appropriately note in the Discussion that reverse causation is plausible (lines 308–317), the framing in the Highlights and Conclusion should use associational language throughout (e.g., "is associated with" rather than "leads to" or "can lead to").
6. Table 1 appears to omit the "Neutral" category for perceived attitude toward older adults. Only "Positive" (26.3%) and "Negative" (28.1%) are listed, leaving 45.5% of the sample unaccounted for in this variable. This category does appear in Table 2a. The authors should either include all three categories in Table 1 or note that the neutral category is omitted for brevity. Additionally, the abbreviations list includes "ANOVA: Analysis of variance," which does not appear to be used anywhere in the study and should be removed.
7. The age distribution of the sample is highly uneven (59.8% aged 60–69, only 8.8% aged 70–79, 31.5% aged 80+), reflecting the GÅS cohort design in which waves 2–4 recruited only the 60- and 81-year-old age groups. While this is a feature of the parent study, it has implications for the generalizability of the age-stratified findings and for the stability of estimates in the 70–79 subgroup. A brief acknowledgment of this distributional imbalance would be appropriate.
8. Finally, the MCS model has a notably low R² of 0.107, indicating that the included covariates explain only about 11% of the variance in mental HRQoL. While this is not uncommon in population studies with broad outcome measures, the authors should comment on this in the Discussion to contextualize the practical significance of the FOC associations observed.
Author Response
International Journal of Environmental Research and Public Health April 24, 2026
Cover Letter
Response to Reviewers
Manuscript ID: ijerph-4195905
Type of manuscript: Original Article
Title: The Association Between Fear of Crime, Life Satisfaction, and Health Related Quality of Life in Non-Victimized Older Adults Aged 60-93 Years. Findings From the Swedish Good Aging in Skåne (GÅS) Population Based Study.
E-mail: henrik.ekstrom@med.lu.se
Dear Editor,
Please find enclosed the revised version of our manuscript with the title “The Association Between Fear of Crime, Life Satisfaction, and Health Related Quality of Life in Non-Victimized Older Adults Aged 60-93 Years. Findings From the Swedish Good Aging in Skåne (GÅS) Population Based Study.”
Manuscript ID: ijerph-4195905
We are grateful for the opportunity to improve the manuscript and we want to thank the reviewers for their fruitful comments and suggestions, which will be responded to by the following point-to-point replies. We hope that the revisions and responses will be sufficient for our manuscript to be accepted for publication in the International Journal of Environmental Research and Public Health.
All authors have approved the revised and re-submitted manuscript.
Henrik Elström
Reviewer 2
This study investigates the association between behavioral fear of crime (FOC) and health-related quality of life (HRQoL) as well as life satisfaction (LS) in a large sample of non-victimized older adults drawn from the Swedish GÅS population study. The topic is timely and clinically relevant, and the exclusion of previously victimized individuals is a meaningful design choice that strengthens the focus on the FOC paradox. The large sample size, population-based design, and the inclusion of chronic disease categories as covariates are notable strengths. However, several methodological and reporting issues require attention before the manuscript is suitable for publication.
Comment 1.
There are multiple numerical inconsistencies in the participant flow description that must be resolved. Line 93 reports that 11,652 individuals were invited across four waves, whereas line 100 states that 11,562 GÅS informants were invited — a discrepancy of 90 individuals that is not explained. Similarly, the wave 3 sample size is reported as n = 2,012 in the text (line 97) but as n = 2,018 in the flow diagram (Figure 1). Most importantly, line 106 states that the study sample comprised 5,832 participants "out of 10,747 eligible individuals," but the preceding arithmetic yields 10,474 eligible individuals after exclusions (11,562 minus the 1,088 who were not contactable, had moved, had language difficulties, or had died). The reported participation percentage of 55.7% is consistent with a denominator of 10,474, not 10,747. These discrepancies undermine confidence in the data pipeline and must be carefully verified and corrected throughout the text and figure.
Authors’ reply
The number and proportions of participants, both in total and in different subgroups, has been corrected in both text and figure.
Comment 2.
The pooling of data from four assessment waves spanning over two decades (2001–2022) without adjusting for wave or calendar period is a substantial methodological concern. Societal factors that plausibly influence both FOC and quality of life — including changes in crime rates, media coverage of violence, urban development, and the COVID-19 pandemic — changed substantially during this period. The authors briefly mention cohort effects in the limitations (lines 370–374), but this acknowledgment does not address the analytical gap. At a minimum, the regression models should include wave or assessment period as a covariate, or a sensitivity analysis should demonstrate that the associations are consistent across waves. Without such adjustment, it is difficult to disentangle the FOC–QoL association from secular trends.
Authors’ reply
To adjust for possible effects of the year in which participants were included, all regression models have been adjusted for the assessment period, waves 1-4. We have not commented further on what effect the different inclusion periods may have had on QoL, we have only noted that the relationship between FOC and QoL remains after adjustment.
Following text has been added in the discussion line 452-457.
“Analyzing historical social or societal causes that can explain FOC in the different cohorts (waves 1-4) such as crime rate, political decisions aimed at reducing crime, newly started associations for the elderly in the area, how the mass media reports on crime or the redevelopment of run-down areas, is beyond the scope of this cross-sectional study, but we can conclude that adjusted for the assessment periods, the associations between FOC and SF-12 PCS, MCS and LSI-A remains.”
Comment 3.
The handling of missing data is not described. The regression model sample sizes differ across outcomes (n = 5,373 for LSI-A, n = 5,367 for PCS, and n = 5,519 for MCS), suggesting that listwise deletion was applied. Given that MADRS has 6.7% missing values and MMSE has 3.8%, the cumulative effect of missing covariates may be non-trivial. The authors should state explicitly how missing data were handled, report the percentage of complete cases used in each model, and discuss whether the missingness pattern could introduce bias. Sensitivity analyses using multiple imputation would strengthen the findings.
Authors’ reply
To check whether the missing data had caused any misinterpretations of the results, we performed sensitivity analyses with multiple imputations. Supplementary tables with original results and results after imputation have been added at the end of the manuscript
The following text has been added, lines 437-442.
“Only participants who had complete data on all variables were included in the regression models, which may lead to the detection of false associations. To check whether the attrition affected the overall result, sensitivity analyses with multiple imputations were performed in all regression models. Comparisons of the results before and after multiple imputations showed minor differences. The overall results remained, see Supplementary tables 1-3, page .
“Only participants who had complete data on all variables were included in the regression models, which may lead to detection of false associations. To check whether the attrition affected the overall result, sensitivity analyses with multiple imputations were performed in all regression models. Comparisons of the results before and after multiple imputations showed minor differences. The overall results remained (Supplementary tables 1-3).” Page 21.”
Comment 4.
The dose-response pattern for FOC and mental HRQoL (MCS) is not monotonic: the "Very often" category (B = −3.17) shows a smaller coefficient than the "Often" category (B = −3.75) in Table 5, which breaks the graded pattern observed in the LSI-A and PCS models. This irregularity deserves explicit discussion. Possible explanations include the small size of the "Very often" subgroup (n = 258, 4.4%), ceiling or floor effects in MCS scoring, or unmeasured characteristics of those with the most extreme avoidance behavior. The authors should not present a uniform dose-response narrative without addressing this deviation.
Authors’ reply
Yes, the grade pattern was broken due to an unmeasured characteristic. In the first analysis we did not include depressive mood as a covariate analyzing MCS. This was a mistake as depressive mood can affect HRQoL and especially MCS in the SF-12, which in turn can reduce any possible association between FOC and MCS. We have included "depressive mood" in the model and obtained a more logical (monotonic) result. We have not commented on this further. See table 5, line 296 and supplementary table 3, line 635-651.
Comment 5.
The Highlights section and the Conclusion contain causal language that is inconsistent with a cross-sectional design. For example, line 17 states that "Fear of crime is a stress factor leading to physical and mental illness," and lines 26–27 and 386 state that "such fear can lead to a reduced quality of life." Although the authors appropriately note in the Discussion that reverse causation is plausible (lines 308–317), the framing in the Highlights and Conclusion should use associational language throughout (e.g., "is associated with" rather than "leads to" or "can lead to").
Authors’ reply
Corrections done to avoid causal language
In highlights “Public health significance” line17. it now says,
“ Fear of crime is a stress factor which may be associated with leading to physical and mental illness.”
In highlights “Public health implications…” line 28, it now says “Politicians and decision-makers at both national and municipal level should be informed about the extent of older adults’ fear of becoming a crime victim and that such fear may impact quality of life negatively. In addition to taking measures against crime itself, the design of the urban environment is important for creating safety.”
Correction in “Conclusion”, line 471
“From a societal perspective, politicians and decision-makers at both national and municipal level should be informed about the extent of FOC among older people and that such fear may be associated with reduced QoL”
Comment 6.
Table 1 appears to omit the "Neutral" category for perceived attitude toward older adults. Only "Positive" (26.3%) and "Negative" (28.1%) are listed, leaving 45.5% of the sample unaccounted for in this variable. This category does appear in Table 2a. The authors should either include all three categories in Table 1 or note that the neutral category is omitted for brevity. Additionally, the abbreviations list includes "ANOVA: Analysis of variance," which does not appear to be used anywhere in the study and should be removed.
Authors’ reply
All three categories for perceived attitude toward older adults are now included in Table 1, line 241. Perceived attitude towards older adults: Positive 1,534 (26.3%), Neutral 1534 (45.5%), Negative 1,639 (28.1%). “ANOVA” is now removed from the list of abbreviation list.
Comment 7.
The age distribution of the sample is highly uneven (59.8% aged 60–69, only 8.8% aged 70–79, 31.5% aged 80+), reflecting the GÅS cohort design in which waves 2–4 recruited only the 60- and 81-year-old age groups. While this is a feature of the parent study, it has implications for the generalizability of the age-stratified findings and for the stability of estimates in the 70–79 subgroup. A brief acknowledgment of this distributional imbalance would be appropriate.
Authors’ reply
We recognize that external validity may be compromised by the uneven age distribution. We have commented on this briefly.
“It should also be mentioned that the uneven age distribution with only 8.8% participants aged 70-79, is another shortcoming that might reduce the generalizability of this study..” Line 412-414.
Comment 8.
Finally, the MCS model has a notably low R² of 0.107, indicating that the included covariates explain only about 11% of the variance in mental HRQoL. While this is not uncommon in population studies with broad outcome measures, the authors should comment on this in the Discussion to contextualize the practical significance of the FOC.
Authors’ reply
Thanks for pointing this out, we have added the following sentences to the discussion line 419-424.
At the same time, despite several relevant explanatory variables being included in the regression models, the R2 value (the coefficient of determination representing the proportion of variance in a dependent variable explained by the independent variables) is small in all regression models, especially in the model with SF-12 MCS as the outcome variable. In practice, this can be interpreted as FOC having a modest significance in explaining the association to life satisfaction or health-related quality of life in older adults.”
Reviewer 3 Report
Comments and Suggestions for Authors
The manuscript addresses an interesting and socially relevant question, namely the relationship between fear of crime, life satisfaction, and health-related quality of life among older adults who have not previously experienced victimization. The topic is important in aging societies, where perceptions of safety may increasingly shape everyday well-being. The study also benefits from a large population-based sample drawn from the Swedish Good Aging in Skåne study, which is clearly a strength. The possibility of adjusting the analyses for several demographic and health-related variables adds further value to the dataset.
At the same time, some aspects of the manuscript could be clarified to strengthen its contribution. The introduction provides a useful overview of fear of crime, but the conceptual link between fear of crime and the outcomes studied is not fully developed. Fear of crime is described as a multidimensional construct, yet the study focuses only on the behavioral dimension. It would help readers if the authors explained more explicitly why this component is expected to be particularly relevant for life satisfaction and HRQoL.
A related point concerns measurement. Fear of crime is assessed with a single item referring to avoiding going out in the evening because of fear of assault or robbery. While this captures an important behavioral expression of fear, it may also reflect other factors, such as mobility limitations, lifestyle patterns, or characteristics of the neighborhood environment. Expanding the discussion of this limitation would improve the methodological transparency of the paper.
The large sample size is an important strength. However, participants were recruited over a long time span (2001–2022), and perceptions of crime and safety can change over time. This issue is mentioned briefly, but it might deserve a slightly fuller reflection in the limitations section.
The statistical analyses appear appropriate, and the associations between higher fear of crime and lower life satisfaction and HRQoL are clearly reported. Still, it would be helpful if the discussion reflected more on the magnitude and practical meaning of these associations. In addition, given the cross-sectional design, the results should be interpreted cautiously. Poor health or lower life satisfaction could also contribute to higher perceived vulnerability and fear of crime.
Overall, the manuscript presents valuable data on an underexplored aspect of well-being in older adults. With a clearer explanation of the conceptual framework, a slightly deeper discussion of measurement issues, and a more cautious interpretation of the findings, the paper could make a useful contribution to research on aging and perceived safety.
Author Response
International Journal of Environmental Research and Public Health April 24, 2026
Cover Letter
Response to Reviewers
Manuscript ID: ijerph-4195905
Type of manuscript: Original Article
Title: The Association Between Fear of Crime, Life Satisfaction, and Health Related Quality of Life in Non-Victimized Older Adults Aged 60-93 Years. Findings From the Swedish Good Aging in Skåne (GÅS) Population Based Study.
E-mail: henrik.ekstrom@med.lu.se
Dear Editor,
Please find enclosed the revised version of our manuscript with the title “The Association Between Fear of Crime, Life Satisfaction, and Health Related Quality of Life in Non-Victimized Older Adults Aged 60-93 Years. Findings From the Swedish Good Aging in Skåne (GÅS) Population Based Study.”
Manuscript ID: ijerph-4195905
We are grateful for the opportunity to improve the manuscript and we want to thank the reviewers for their fruitful comments and suggestions, which will be responded to by the following point-to-point replies. We hope that the revisions and responses will be sufficient for our manuscript to be accepted for publication in the International Journal of Environmental Research and Public Health.
All authors have approved the revised and re-submitted manuscript.
Henrik Ekström
Reviewer 3
The manuscript addresses an interesting and socially relevant question, namely the relationship between fear of crime, life satisfaction, and health-related quality of life among older adults who have not previously experienced victimization. The topic is important in aging societies, where perceptions of safety may increasingly shape everyday well-being. The study also benefits from a large population-based sample drawn from the Swedish Good Aging in Skåne study, which is clearly a strength. The possibility of adjusting the analyses for several demographic and health-related variables adds further value to the dataset.
Comment 1.
At the same time, some aspects of the manuscript could be clarified to strengthen its contribution. The introduction provides a useful overview of fear of crime, but the conceptual link between fear of crime and the outcomes studied is not fully developed. Fear of crime is described as a multidimensional construct, yet the study focuses only on the behavioral dimension. It would help readers if the authors explained more explicitly why this component is expected to be particularly relevant for life satisfaction and HRQoL.
Authors’ reply
We have clarified the connection between behavioral FOC and QoL by adding the following text in the introduction, line 86-93: “The behavioral aspect of FOC influences how people interact with their environment, their communities, and may affect everyday life. It may become more difficult to implement collective solutions and collaborations within the local community. Not daring to go out in the evening can, at an individual level, lead to an increased degree of loneliness and social isolation and a risk of addictive behavior. If the behavior means that one feels compelled to protect oneself by installing surveillance cameras or alarms, it can entail financial costs. Not daring to be outdoors can lead to poorer health, either because one is less physically active or because one avoids seeking care.”
Comment 2.
A related point concerns measurement. Fear of crime is assessed with a single item referring to avoiding going out in the evening because of fear of assault or robbery. While this captures an important behavioral expression of fear, it may also reflect other factors, such as mobility limitations, lifestyle patterns, or characteristics of the neighborhood environment. Expanding the discussion of this limitation would improve the methodological transparency of the paper.
Authors’ reply
We have tried to improve the transparency by adding your suggestions (thank you!) with the text:
“Not daring to go outdoors due to fear of becoming a victim of crime was the starting point of the study. At the same time, it should be remembered that there are many more reasons why people avoid going out. We have taken some into account in our analysis, such as ADL, cognition, depressive mood and illness. Important reasons that have not been included are, for example, characteristics of the neighborhood environment or mobility limitations.” Line 443-448.
Comment 3.
The large sample size is an important strength. However, participants were recruited over a long time span (2001–2022), and perceptions of crime and safety can change over time. This issue is mentioned briefly, but it might deserve a slightly fuller reflection in the limitations section.
Authors’ reply
Thank you for pointing this out. We are well aware of the time factor and that FOC can change overtime for a number of reasons. At the same time, we do not have data available to include any historical analyses in this cross-sectional study. To explain this, we have added the following text, line 452-457.
“Analyzing historical social or societal causes that can explain FOC in the different cohorts (waves 1-4) such as crime rate, political decisions aimed at reducing crime, newly started associations for the elderly in the area, how the mass media reports on crime or the redevelopment of run-down areas, is beyond the scope of this cross-sectional study, but we can conclude that adjusted for the assessment periods, the associations between FOC and SF-12 PCS, MCS and LSI-A remains.”
Comment 4 a.
The statistical analyses appear appropriate, and the associations between higher fear of crime and lower life satisfaction and HRQoL are clearly reported. Still, it would be helpful if the discussion reflected more on the magnitude and practical meaning of these associations.
Authors’ reply
The magnitude and practical meaning of the associations between FOC and QoL has been commented on in the second paragraph of the discussion with the following text:
“Although we have shown statistical relationships between FOC and QoL, the question is whether these are relevant and whether the results mean anything in practice. In general, it can be said that it depends on what the study population looks like and in what context the survey is conducted. However, if one looks at life satisfaction as a global measure of well-being, the norm values for a general Swedish population 60 years and older have been calculated for the LSA-A to be 27.8 points with a standard deviation of 6.9 points. The adjusted difference of 2 points, between the groups of never being influenced by FOC in their behavior to very often doing so, is not insignificant [18]. In addition to a worse mood and deteriorating mental and physical health and a reduced social engagement of the individual [28], lower life satisfaction can indirectly lead to a higher FOC (the inverse relationship we have tried to highlight) and possibly contribute to negative societal consequences, for example an increased burden on healthcare and social care both in terms of work effort and costs [29].” Line 309-322.
Comment 4 b.
In addition, given the cross-sectional design, the results should be interpreted cautiously. Poor health or lower life satisfaction could also contribute to higher perceived vulnerability and fear of crime.
Authors’ reply
We have removed all text that suggests causality and only talk about possible associations. We have also commented that lower life satisfaction could also contribute to higher perceived vulnerability and fear of crime with the text:
“The reverse relationship is plausible, i.e., that lower QoL leads to greater FOC. People who for several reasons, including medical, financial, or social factors, may perceive their QoL to be impaired might consider themselves more vulnerable and therefore have a greater fear of being exposed to violence or threats of violence [39].” Line 361-364.
Additionally, we have added the following lines:
“However, it should be said that the cross-sectional design of this study precludes the evaluation of the direction of the associations of FOC and LS and HRQoL.” Line 425-426
Comment 5
Overall, the manuscript presents valuable data on an underexplored aspect of well-being in older adults. With a clearer explanation of the conceptual framework, a slightly deeper discussion of measurement issues, and a more cautious interpretation of the findings, the paper could make a useful contribution to research on aging and perceived safety.
Authors’ reply
We are grateful for all the time and effort you put into reviewing the manuscript.
Round 2
Reviewer 1 Report
Comments and Suggestions for Authors
Thank you for your thorough response to my comments.
However, I remain somewhat unconvinced by the reluctance to report confidence intervals, especially since their calculation is standard in this type of analysis. I would like to gently reiterate that reporting confidence intervals alongside odds ratios is common practice precisely because it improves statistical interpretability and allows readers to assess the precision of estimates more directly.
Author Response
Dear Reviewer,
I completely agree that odds ratios should be presented with confidence intervals,
and with all due respect to your comment, I see nowhere in the manuscript where the odds
ratios are not presented with confidence intervals. Again, thank you for all the work
and time you put into reviewing the manuscript.
Best regards
Henrik Ekström
Reviewer 2 Report
Comments and Suggestions for Authors
I thank the author for responding to and revising my comments; my questions have now been resolved, and I recommend the publication of this paper.
Author Response
Dear reviewer,
Once again, thank you for all the work and time you put into reviewing the manuscript.
It has been very valuable to us.
Best regards
Henrik Ekström
